Commission on Inequality in smf.co.uk/education

Transcription

1 Educational inequalities in England and Wales Commission on Inequality in Education

2 Contents Introduction and summary findings Findings for secondary school: GCSE results - Inequalities by income, gender, ethnicity and region - How inequalities have evolved over the past decades Findings for primary school: age 11 results - Inequalities by income, gender, ethnicity and region - How inequalities have evolved over the past decades

3 Introduction and summary findings

4 Introduction and summary This slide pack sets out our findings on inequalities in educational attainment at age 16 and age 11. It describes four aspects of inequality and how these affect school results: region, family income, gender and ethnicity. We also examine how these trends in inequality have evolved over time, looking at cohorts of children born in different years: 1958, 1970, 1997/98 and 2000.

5 Headline findings 1: Inequalities by place Regional inequalities GCSE performance at age 16 across England and Wales reveals marked disparities between regions, with over 70% of pupils in London achieving 5 good GCSEs compared to 63% in Yorkshire & Humber. Regional differences in attainment are already apparent by the end of primary school and they are observable even when you control for other factors such as ethnicity and income. Our analysis across different cohorts of children sitting exams at age 16 shows that regional inequalities have remained stubborn and in some cases worsened over the last three decades. Areas such as the North East, Yorkshire and the Humber, the West Midlands and the East Midlands have persistently under-performed, behind whilst London s performance has surged. Comparing the performance of 11-year olds born in 2000 with those born in 1970 reveals that the geographic area a child comes from has become a more powerful predictive factor for those born in 2000 compared to 1970.

6 Headline findings 2: Inequalities by ethnicity and income Inequalities by family income A very low proportion of pupils who receive Free School Meals achieve 5 A* to C at GCSEs (40%) compared to those not receiving Free School Meals (70%). The performance gap between the richest and the poorest has remained persistently large between the mid-1980s and the mid-2000s. When using the measure of 5 A* to C grades, the attainment gap between FSM pupils and non-fsm pupils is observed to narrow over the last decade. However, when more demanding measures, such as 5 A* to C including English and Maths are applied, this progress is no-longer observable. Inequalities by ethnic group Educational performance varies significantly across different ethnic minority groups. For instance, while over 85% of Chinese pupils get five good GCSEs, only around 59% of Black Caribbean pupils achieve this benchmark. Ethnic differences are important in their own right as distinct from income, region or other factors with the performance of a Chinese child at age 11 higher than for a white child of the same age. Over the last three decades, ethnic inequalities have altered radically but a similar level of unevenness remains. While Asian students born in 1970 performed poorly, Chinese, Indian and Bangladeshi born in 1997/98 were the best performers. White students have fallen from over-performers to under-performers on average over the three decades.

7 Secondary school: GCSE results

8 Secondary school: GCSE results What we did Our analysis of GCSE results is based on Department for Education statistics, broken down by income, gender, ethnicity and region. These statistics are based on exam results of all students taking GCSEs in England and Wales. To look at how inequalities have evolved over time, we use data from the British Cohort Study (BCS) covering children born in This follows the lives of 17,000 children born in 1970, and contains data on their O-level results, taken in

9 Secondary school: GCSE results London and the South out-perform the rest of the country Attainment by region, Percentage achieving 5+ A*-C GCSEs or equivalent 75.0% 70.0% 65.0% Big differences in GCSE results between worst performing and best preforming areas. 60.0% 55.0% 50.0% Yorkshire and the East Midlands North East West Midlands North West East South West South East Inner London London Outer London Over 70% of pupils in London achieve five good GCSEs compared to 63% in Yorkshire & Humber Source: Department for Education

10 Secondary school: GCSE results Poorer students are less likely to receive five good GCSEs Attainment by Free School Meal status, Percentage achieving 5+ A*-C GCSEs or equivalent 80.0% 70.0% 60.0% 50.0% 40.0% Pupils are eligible for free school meals if their families receive certain income-related benefits. As such, it provides a proxy for pupils from a low income background. 30.0% 20.0% 10.0% 0.0% FSM non-fsm Just over 40% of pupils who are eligible for free school meals get five good GCSEs, compared to almost 70% of pupils who are not eligible for free school meals. Source: Department for Education

11 Secondary school: GCSE results The picture for income inequality is similar for Wales 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Attainment by Free School Meal status, Wales, Level 2 Level 2 including English/Welsh and Maths Differences between England and Wales mean it is difficult to compare results across the two directly. However, it is clear that the gap between poorer and richer pupils in Wales is of a similar magnitude to that in England. Level 2 is defined as 5 A*-C GCSEs or equivalent Source: Statistics for Wales

12 Secondary school: GCSE results Girls out-perform boys 80.00% Attainment by gender, Percentage achieving 5+ A*-C GCSEs or equivalent 70.00% 60.00% 50.00% Over 71% of girls achieve five good GCSEs compared with just under 60% of boys 40.00% 30.00% 20.00% 10.00% 0.00% Boys Girls Source: Department for Education

13 Secondary school: GCSE results Chinese and Indian groups out-perform; Black Caribbean and Pakistani underperform 90.0% 85.0% 80.0% 75.0% 70.0% 65.0% 60.0% 55.0% 50.0% Attainment by ethnicity, Percentage achieving 5+ A*-C GCSEs or equivalent White Mixed Indian Pakistani Bangladeshi Black Caribbean Black African Chinese Source: Department for Education There are substantial differences across ethnicity. Some ethnic minority groups out-perform by a large margin: over 85% of Chinese pupils get five good GCSEs. But others lag behind. Only around 59% of Black Caribbean pupils get five good GCSEs

14 Secondary school: GCSE results Secondary school: GCSE results Our analysis of the latest data on GCSE performance at age 16 across England and Wales illustrates the significant inequalities that exist across region, family income, ethnicity and gender. Particularly noteworthy is the significant disparity between regions with over 70% of pupils in London achieve good GCSEs compared to 63% in Yorkshire & Humber. In the next section, we look at how this has changed over time to understand whether there have been improvements and where inequalities may have worsened. To do this, we compare performance in use data from the 1970 British Cohort Study (BCS), which follows the lives of 17,000 children born in 1970, and contains data on their O-level results, taken in We do this rather than use official examination results for all pupils because the BCS is a rich source of data that allows us to break down results by family background, region and gender.

15 Secondary school: changing pattern of inequalities Little progress in evening out regional inequalities over three decades, with some areas falling further behind The East Midlands, West Midlands, North East and Yorkshire & Humberside have persistently under-performed compared to the national average. The South-East maintained its higher than average performance. London has seen an especially dramatic improvement cohort taking O-levels in Percentage achieving 5+ A-C O-levels results: Percentage achieving 5+ A*-C GCSEs or equivalent South East East of England Wales London South West North West Yorkshire and North East West Midlands East Midlands 0% 10% 20% 30% 40% Sourc e: SMF Analy s is of Britis h Cohort Study London South East South West East of England North West West Midlands North East East Midlands Yorkshire and 0% 20% 40% 60% 80%

16 Persistently large gap between richest and poorest Secondary school: changing pattern of inequalities Attainment by birth year Percentage achieving 5+ A*-C O-levels/GCSEs or equivalent 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% Previous research by Blanden and Macmillian (2014) comparing cohorts of individuals born in 1958, 1970 and 1989/90 shows that there are persistent, longstanding gaps in performance between children from the poorest and richest families. 20.0% 10.0% 0.0% /1990 Most Deprived 20% Least Deprived 20% Source: Blanden and Macmillan (2014)

17 Secondary school: changing pattern of inequalities In more recent years, much of the apparent narrowing of the gap between rich and poor disappears when more demanding measures are used Attainment by FSM status Percentage achieving 5+ A*-C GCSEs or equivalent 2009/ / / / /2014 ( New Methodology) In recent years, there was an apparent narrowing of the gap in performance between pupils on free schools meals and other pupils. However, this mainly applied to the 5 A*-C or equivalents measure, not to 5 A*-C including English and Maths (Lupton and Thomson, 2015). In , a new methodology was brought in to measure performance. The new methodology is more restrictive in what qualifications are counted as equivalent to GCSEs. On this new measure, gap between FSM and non-fsm pupils is much larger. FSM Non-FSM Source: Department for Education

18 Secondary school: changing pattern of inequalities The gender gap has widened over time Attainment by gender Percentage achieving 5+ A*-C O-levels/GCSEs or equivalent 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% Qualifications and performance measures have changed over the past decades, making direct comparisons difficult. However, the gender gap with girls out-performing boys appears to have become starker over time 0.0% Born in 1970, taking O- levels in Born in , taking GCSEs in Male Female Source: SMF Analysis of British Cohort Study; Department for Education

19 Secondary school: changing pattern of inequalities Some ethnic minorities are doing much better than previously Children from some ethnic groups are now doing much better than average It is hard to compare against past performance as ethnicity was not captured in the same way; however the large penalty previously associated with Asian appears to have fallen dramatically. Chinese and Indian students outperform the average by a large margin cohort taking O-levels in Percentage point difference between the proportion achieving 5+ A-C O-levels and the population average results: Percentage point difference between the proportion achieving 5+ A*-C GCSEs or equivalent and the population average 2% 0% -2% -4% -6% -8% -10% -12% -14% 30% 20% 10% 0% -10% White Mixed Indian Pakistani Bangladeshi Black Caribbean White Others Asian Black African Chinese Source: SMF Analysis of British Cohort Study; Department for Education

20 Secondary school: conclusions Conclusions from the latest GCSE results Marked disparities between regions, with over 70% of pupils in London achieving 5 good GCSEs compared to 63% in Yorkshire & Humber. Low proportion of pupils who receive Free School Meals achieving 5 A* to C at GCSEs (40%) compared to those who do not receive Free School Meals (70%). Attainment varies across different ethnic minority groups. While over 85% of Chinese pupils get five good GCSEs, only around 59% of Black Caribbean pupils achieve this benchmark. Conclusions on changing pattern of inequalities over time Regional disparities persist, with some areas such as the North East, Yorkshire and the Humber and the East Midlands falling further behind and London s performance surging over the last three decades. The performance gap between the richest and the poorest has remained persistently large between the mid-1980s and the mid-2000s. When using the measure of 5 A* to C grades, the attainment gap between FSM pupils and non-fsm pupils is observed to narrow over the last decade. However, when more demanding measures are applied, this progress is no-longer observable. Ethnic inequalities have altered radically but have not levelled out. While Asian students born in 1970 performed poorly, Chinese, Indian and Bangladeshi born in 1997/98 were the best performers. White students have fallen from over-performers to under-performers on average over the three decades. Between 1970 and 2013/2014, girls have advanced their attainment compared to boys.

21 Primary school: age 11 results

22 Primary school: Age 11 results What we did Our analysis for age 11 focuses on evidence from cognitive test scores from two different cohort studies: Children born in 1970 (British Cohort Study or BCS) Children born in 2000 (Millennium Cohort Study, or MCS) We used the results from verbal reasoning tests. The rationale for doing this was threefold: These tests are comparable over time (unlike KS2 tests which were introduced more recently, and other cognitive test scores which are not consistently available). Evidence suggests that verbal reasoning is an important indicator of later attainment (Strand 2006). Analysis of attainment levels in verbal reasoning tests provides a complementary method of assessing inequalities. It strips out the potential effect that grade inflation in public exams may have on attainment and the inequalities that we observe. We examine inequality across the following dimensions: region, income and ethnicity. We do not report results for gender as the evidence suggests that the performance of boys and girls may vary depending on the type of test.

23 Primary school: Age 11 results Regional differences are already apparent at primary school level To analyse the data, we have split pupils into three groups depending on their test results: low ability, mid ability and high ability. The table below shows the distribution of each group across regions, and compared to the survey population. Yorkshire & Humberside and the West Midlands have disproportionately high numbers of lowscoring pupils. The North West, London and Wales have disproportionately high numbers of high-scoring pupils. Note: the survey populations for the MCS dataset are not the same as the national average as the sample was boosted in certain areas. Survey population Low ability group Mid ability group High ability group North East 4% 4% 3% 3% North West 11% 9% 9% 14% Yorkshire & Humber 9% 12% 9% 5% East Midlands 7% 7% 8% 7% West Midlands 8% 9% 7% 7% East of England 9% 9% 11% 8% London 10% 9% 10% 13% South East 15% 15% 17% 14% South West 9% 9% 9% 7% Wales 18% 17% 17% 21% 23 Source: SMF Analysis of Millennium Cohort Study

24 Primary school: Age 11 results High ability children are more likely to come from the top income deciles As before, we have split pupils into three groups depending on their test results: low ability, mid ability and high ability. The table below shows the distribution of each group across family income deciles, where decile 1 is made up of the poorest households, and decile 10 is made up of the richest. A disproportionately high number 14% of high scoring pupils - come from the richest 10% of households. Meanwhile 17% of low scoring pupils come from the poorest 10% of families. Income decile Survey population Low ability group Mid ability group High ability group 1 Poorest 10% 17% 8% 5% 2 10% 15% 9% 6% 3 10% 12% 10% 8% 4 10% 11% 10% 8% 5 10% 9% 10% 11% 6 10% 9% 11% 10% 7 10% 8% 10% 12% 8 10% 7% 11% 13% 9 10% 6% 11% 12% 10 Richest 10% 5% 10% 14% Source: SMF Analysis of Millennium Cohort Study

25 Primary school: Age 11 results Several ethnic minority groups out-perform the average, but many lag behind As before, we have split pupils into three groups depending on their test results: low ability, mid ability and high ability. The table below shows the distribution of each group across ethnic groups. The Chinese, Indian, Black African and Other Asian groups have disproportionately high numbers of high scoring pupils. The Pakistani, Bangladeshi, Black Caribbean and other Black groups have disproportionately high numbers of low scoring pupils. Note: these are figures for the MCS sample, which was boosted for certain ethnic groups therefore figures not necessarily representative of the national population. Ethnic group Survey population Low ability group Mid ability group High ability group White 88.1% 86.1% 90.0% 88.9% Mixed 2.5% 2.2% 2.7% 2.6% Indian 1.7% 1.4% 1.6% 2.3% Pakistani 2.8% 4.5% 2.1% 1.9% Bangladeshi 1.2% 2.2% 0.6% 0.6% Other Asian 0.6% 0.6% 0.5% 0.7% Black Caribbean 0.7% 0.8% 0.7% 0.7% Black African 1.4% 1.3% 1.4% 1.5% Other Black 0.2% 0.3% 0.1% 0.2% Chinese 0.1% 0.0% 0.0% 0.2% Any other 0.5% 0.7% 0.4% % Source: SMF Analysis of Millennium Cohort Study

26 Primary school: Age 11 results But these aspects of difference are related to each other There are a number of inter-relationships between ethnicity, income and region. For example, Black Caribbean children are, on average, in lower income households than Indian children. Some regions have higher average incomes, and some regions have higher proportions of certain ethnic groups. In the following slides, we present the results of regression analysis which explores which effects ethnicity, income or region are strongest once other factors are taken into account. In the following charts, the verbal reasoning test scores are standardised: mean 0, standard deviation 1. This means: a score of 0 represents the average score; a score of 1 loosely represents an average high ability score; a score of -1 loosely represents an average low ability score

27 Income, ethnicity and region are all important drivers of test scores Being in the top income decile rather than the bottom income decile increases a child s expected score by about 0.9 so takes the expected score from being average to being high ability. This is after taking into account other effects such as region and ethnicity. Doubling family income adds around 0.3 to the expected score. Primary school: Age 11 results Overall predictors of verbal reasoning score: children born in Top decile rather than bottom decile Top quintile rather than bottom quintile Chinese vs white Doubling family income Indian vs white Black African vs white NW vs London Female West Midlands vs London Bangladeshi vs white Yorkshire vs London Source: SMF Analysis of Millennium Cohort Study

28 Primary school: changing pattern of inequalities For children born in 1970, income was important, but region less so For children born in 1970, doubling family income had a stronger effect adding 0.5 rather than 0.3 to child s test score. However, being in top decile rather than the bottom decile was slightly less important adding 0.8 rather than 0.9 to a child s expected score. For children born in 1970, non-white ethnic groups did much worse than the white group. Regional differences were very small. Overall predictors of verbal reasoning score: children born in Top decile rather than bottom decile Top quintile rather than bottom quintile Source: SMF Analysis of Millennium Cohort Study Doubling family income SW vs SE Female West Indian vs white Indian vs white Pakistani vs white

29 Primary school: conclusions Findings for 11 year-olds born in 2000 Regional differences in attainment are already apparent by the end of primary school. High ability children at age 11 are much more likely to come from families in the top income deciles and low ability children to come from families in the bottom income deciles. When controlling for all other observable factors, we find that: - Being in the top income decile rather than the bottom income decile increases a child s expected score from being average to being high ability. - Ethnic differences are important in their own right with the performance of a Chinese child higher than for a white child similar in all other ways. - Regional differences matter: for instance, being from London rather than Yorkshire is associated with better attainment all other things being equal. Conclusions on changing pattern of inequalities over time The effect on attainment of doubling the family income was lower for children born in 2000 than those born in 1970 suggesting a flattening of inequalities. However, this masks more marked inequalities at the extremes: Being in the top rather than bottom decile became a stronger predictor of a high score in 2000 than 1970 (adding 0.8 for the 1970 cohort rather than 0.9 for the 2000 cohort) The geographic area a child comes from has become a more powerful predictive factor for those born in 2000 compared to 1970.